Method for error detection by using top-down method

Abstract

Disclosed is a method for error detection, performed by one or more processors of a computing device according to an example embodiment of the present disclosure. The method includes evaluating an error rate for a sentence to be evaluated, in a first language unit. the method includes evaluating an error rate in a second language unit which is smaller than the first language unit, based on the first language unit error.

What is claimed is: 
     
       1. A method for error detection performed by a computing device, comprising:
 evaluating an error rate for a sentence to be evaluated, in a first language unit; 
 identifying a part where a predetermined type of error occurs in the first language unit; and 
 evaluating an error rate for the part where a predetermined type of error occurs, in a second language unit which is smaller than the first language unit. 
 
     
     
       2. The method according to  claim 1 , further comprising:
 evaluating an error rate in a third language unit which is smaller than the second language unit, based on the second language unit error. 
 
     
     
       3. The method according to  claim 2 , wherein the third language unit includes a grapheme unit and the error rate evaluation result of the third language unit includes a grapheme error rate (GER) evaluation result when the second language unit includes the character unit or includes a character error rate (CER) evaluation result when the second language unit includes the sub-word unit. 
     
     
       4. The method according to  claim 2 , wherein the evaluating of an error rate in a third language unit which is smaller than the second language unit, based on the second language unit error includes:
 identifying a part where a predetermined type of error occurs in the second language unit; and 
 evaluating an error rate for the part where a predetermined type of error occurs, in the third language unit. 
 
     
     
       5. The method according to  claim 4 , wherein the identifying of a part where a predetermined type of error occurs in the second language unit includes:
 identifying a character or a sub-word in which the substitution error occurs, in the character unit or the sub-word unit, and 
 the evaluating of an error rate for the part where a predetermined type of error occurs, in the third language unit, includes: 
 evaluating an error rate for the character in which the substitution error occurs, in the grapheme unit when the second language unit includes the character unit; or 
 evaluating an error rate for the sub-word in which the substitution error occurs, in the character unit when the second language unit includes the sub-word unit. 
 
     
     
       6. The method according to  claim 1 , wherein the first language unit includes a word unit and the second language unit includes a character unit or a sub-word unit. 
     
     
       7. The method according to  claim 1 , wherein the error rate evaluation result of the first language unit includes a word error rate (WER) evaluation result and the error rate evaluation result of the second language unit includes a character error rate (CER) evaluation result or a sub-word error rate (SER) evaluation result. 
     
     
       8. The method according to  claim 1  or  2 , wherein the error rate evaluation includes at least one of:
 evaluation of whether substitution occurs, 
 evaluation of whether deletion occurs, and 
 evaluation of whether insertion occurs. 
 
     
     
       9. The method according to  claim 1 , wherein the predetermined type of error includes a substitution error. 
     
     
       10. The method according to  claim 1 , wherein the identifying of a part where a predetermined type of error occurs in the first language unit includes:
 identifying a word in which the substitution error occurs, in the word unit, and 
 the evaluating of an error rate for the part where a predetermined type of error occurs, in the second language unit, includes: 
 evaluating an error rate for the word in which the substitution error occurs, in a character unit or a sub-word unit. 
 
     
     
       11. The method according to  claim 1 , further comprising:
 calculating a final error rate in consideration of an error rate of the first language unit and an error rate of the second language unit. 
 
     
     
       12. A computer program stored in a computer readable storage medium which causes one or more processors to perform operations for detecting an error when the computer program is executed by the one or more processors, wherein the operations include:
 an operation of evaluating an error rate for a sentence to be evaluated, in a first language unit: 
 an operation of identifying a part where a predetermined type of error occurs in the first language unit; and 
 an operation of evaluating an error rate for the part where a predetermined type of error occurs, in a second language unit which is smaller than the first language unit. 
 
     
     
       13. A computing device, comprising:
 at least one processor; and 
 a memory, 
 wherein at least one processor is configured to evaluate an error rate for a sentence to be evaluated, in a first language unit 
 identify a part where a predetermined type of error occurs in the first language unit; and 
 evaluate an error rate for the part where a predetermined type of error occurs, in a second language unit which is smaller than the first language unit.